Pozzebon et al.,2010

More documents

Recommendations

Info

OREGON VOCATIONAL INTEREST SCALES 169 them completed the survey. 1 Over the years, Goldberg has carried out a number of analyses of CISS scales, the most important of which for our purposes were analyses of the seven CISS Orientation scales. To develop public-domain measures of each of the CISS Orientations, 2,035 items from the International Personality Item Pool (IPIP; Goldberg, 1999) were correlated with the seven scale scores, and IPIP items were classified by the CISS scale with which they were most highly associated. IPIP items falling within each category were then selected rationally based on the extent of their correlations with the CISS scale, the apparent relevance of their content to the construct, and their lack of redundancy with other items already selected for that IPIP scale. Finally, the reliability of preliminary versions of the new scales were analyzed, and any items that served to attenuate scale reliability were omitted and in some cases replaced with other IPIP items that functioned more adequately. All IPIP items are short phrases, beginning with a verb (e.g., “Take risks,” “Talk softly”). Those IPIP items that turned out to be most highly associated with the CISS scales typically included verbal phrases involving interest or preference (e.g., “Like,” “Do not like,” “Enjoy,” “Do not enjoy,” “Prefer,” “Am [not] interested in”). To discover whether the self-reported relative frequencies of individuals’ actual behavioral acts might turn out to be even better measures of interests, Goldberg used the 400 items in the Behavioral Report Inventory (BRI), which had been administered to the ESCS in 1997, to develop BRI scales associated with the seven CISS Orientation scales using the exact same procedures used to develop the IPIP scales. Both the IPIP and BRI scales were developed in parallel in 2004 and compared as predictors of the CISS constructs. Information about both sets of scales is available from L. R. Goldberg. In factor analyses of the original CISS Orientation scales, separately for skills and interests, as well as for the new IPIP and BRI versions of those seven constructs, it was always necessary to extract eight factors for the seven scales to each load most highly on a separate factor. If less than eight factors were extracted, the scales measuring the Producing and Analyzing Orientations always loaded most highly on the same factor. In the eight-factor analyses of the original CISS scales, the additional factor included CISS scales measuring interests and skills related to such occupations as Translator/Interpreter, Writer/Editor, Librarian, Liberal Arts Professor, and Musician and to such basic interests as Writing and International Activities. Seemingly, then, the addition of an eighth dimension (which Goldberg called “Erudition”) to the Holland (1973) six and the Campbell et al. (1992) seven might be warranted. Although such an additional dimension is unlikely to be completely independent of the other constructs, it might serve to capture important individual differences unavailable in previous vocational inventories. The ORVIS were developed as direct measures of these eight constructs. Items for each of the eight new scales were generated by Goldberg to include both interests and activities that were conceptually associated with each dimension, based on the corresponding IPIP and BRI scales, and on the content of the CISS 1 Of the more than 30 questionnaires administered to the ESCS during the 1993 to 2006 period, the CISS was the only one for which no honorarium payment was provided; and doubtless because of this, participation was lower than for the other surveys scales most highly associated with the additional eighth factor. All items from the preliminary versions of these new ORVIS scales were administered to the ESCS as part of an Omnibus Personal Attributes Survey (OPAS) in 2006. This article is the first report of our analyses of these scales in this community sample along with a cross-validation of our findings in another quite different kind of sample: one consisting of college students. Thus, this article represents the first empirical validation of this construct-driven measure of vocational interests. METHOD Participants Community sample. From Goldberg’s (1999) ESCS, 665 participants completed the ORVIS, of whom 379 (57%) were women and 286 (43%) were men. In 2006, the participants’ mean age was 62 years (SD = 11.7). Over 98% of respondents were White, and 85% had at least some college education. College sample. Canadian college students in their 1st year of study participated in an ongoing longitudinal investigation of academic performance, college satisfaction, and choice of major. Of the 346 students, there were 245 women (71%) and 101 men (29%), with a mean age of 18.5 years (SD = 1.7). Measures ORVIS. Participants in both samples completed the 92 ORVIS items (see the Appendix for items sorted by scale). For each item, participants rated their level of interest in each occupational description (e.g., “Care for sick people”) on a 5-point scale ranging from 1 (strongly dislike)to5(strongly like). CISS. Most participants of the community sample also completed the CISS (Campbell et al., 1992), which contains 320 self-report items, each using a 6-point response scale ranging from 1 (strongly dislike) to6(strongly like). We used participants’ scores on the seven CISS Orientation scales: Influencing, Organizing, Helping, Creating, Analyzing, Producing, and Adventuring. These scales correspond conceptually to the first seven ORVIS scales listed previously. (There is no direct counterpart of the ORVIS Erudition scale in the CISS, but much of the content of ORVIS Erudition is represented within CISS Creating.) Personality. Participants in the community sample completed various personality measures including the IPIP Big Five scales (Goldberg, 1999). Internal consistency reliabilities (alpha) of the IPIP Big Five scales ranged from .88 to .91 in this sample. Participants in both samples also provided self-reports on measures of the HEXACO personality factors. Specifically, community sample participants provided self-reports on the fulllength (192-item) form of the HEXACO Personality Inventory (HEXACO–PI; Lee & Ashton, 2004), whereas college sample participants provided self-reports on the half-length (100item) form of the same inventory (HEXACO–PI–R; e.g., Lee et al., 2009). Both versions of the HEXACO inventory assess six broad personality factors: Honesty-Humility, Emotionality, Extraversion, Agreeableness, Conscientiousness, and Openness to Experience. In the short version of the inventory, one of the four facet-level scales defining the Extraversion factor has been replaced. For consistency across samples, we computed scores on the Extraversion factor from the three common facet-level
170 POZZEBON ET AL. scales only, and we computed scale scores from the items of the half-length form. All items were administered using a response scale with options ranging from 1 (strongly disagree)to5 (strongly agree). Internal consistency reliabilities (alpha) of the six HEXACO scales ranged from .79 to .84 in the community sample and from .81 to .84 in the college sample. Cognitive ability. In the community sample, we measured reasoning ability was measured by Factor B, a 15-item Reasoning scale, from the Sixteen Personality Factor Questionnaire (16PF; Conn & Rieke, 1994). Although the 16PF is a personality inventory, the items of its Reasoning scale are cognitive ability items, scored as correct or incorrect. The internal-consistency reliability of the Reasoning scale was .75 in this sample. In the college sample, we administered two measures of ability. Verbal ability was measured with the 46 items of the Vocabulary scale from the Multidimensional Aptitude Battery (MAB; Jackson, 1984). Mathematical ability was measured with a test consisting of 20 items taken from the Gauss Mathematics Contest (Center for Education in Mathematics and Computing, 2007); the mathematics items assessed problem solving in arithmetic and in basic algebra and geometry. Internal-consistency reliabilities of the verbal and mathematical ability measures were .78 and .65, respectively. RESULTS Internal Consistency Reliabilities Internal consistency reliabilities and descriptive statistics for the ORVIS scales are reported in Table 1 for both the community and college samples. The reliabilities were generally high in both samples, ranging from .79 and .75 for Erudition to .90 and .91 for Organization, respectively. As shown in Table 1, the means for all scales in both samples were reasonably close to the theoretical midpoints (i.e., 3.00 for the 1–5 scales), and the standard deviations were reasonably wide, covering at least one fifth of the theoretical range. The mean scores for women and men indicated some sex differences in the scales. In both samples, the largest sex difference was in Adventure, with men’s scores more than a standard deviation higher than women’s scores. Men’s Analysis scores were also substantially higher than were women’s (d >.5), whereas women scored higher on Altruism than did men (d >.5). Factor Structure of the ORVIS Items We also conducted item-level factor analyses of the 92 ORVIS items in each of the two samples. In both samples, we extracted eight principal components and rotated them to an orthogonal Procrustes solution based on a target matrix in which each item was assigned a target loading of 1 for its designated scale and 0 for all other scales (see the Appendix for the loading of each item on its targeted factor). In both samples, factor scores on the resulting factors correlated strongly with the corresponding scale scores: Correlations ranged from .69 (Erudition) to .94 (Altruism) in the college sample and from .77 (Erudition) to .94 (Organization and Altruism) in the community sample. Thus, the factor analysis results generally supported the division of the ORVIS items into the eight specified scales. 2 2 A confirmatory factor analysis of the ORVIS items is unsuitable for the following reasons. First, the ORVIS scales will not be strictly unidimensional, TABLE 1.—Means, standard deviations, and sex differences in the Oregon Vocational Interest Scales. Total Women Men Scale M SD M SD M SD d(Women–Men) Community sample Leadership (.87) 2.78 .76 2.66 .75 2.94 .75 −.37 Organization (.90) 2.52 .79 2.48 .80 2.58 .77 −.12 ∗ Altruism (.86) 3.15 .70 3.32 .65 2.93 .70 .60 Creativity (.88) 3.22 .80 3.39 .81 2.99 .73 .53 Analysis (.88) 2.60 .87 2.38 .82 2.90 .84 −.62 Production (.81) 3.17 .72 3.05 .71 3.34 .71 −.40 Adventure (.82) 2.50 .77 2.20 .66 2.89 .73 −1.03 Erudition (.79) 3.58 .69 3.40 .70 3.12 .65 .41 College sample Leadership (.85) 2.90 .74 2.82 .74 3.10 .70 −.36 Organization (.91) 2.38 .81 2.29 .77 2.58 .87 −.45 Altruism (.84) 3.24 .70 3.40 .64 2.87 .69 .73 Creativity (.89) 2.96 .86 3.01 .86 2.84 .86 .18 ∗ Analysis (.84) 2.08 .73 1.94 .64 2.42 .81 −.86 Production (.82) 2.33 .73 2.22 .68 2.60 .76 −.49 Adventure (.83) 2.78 .82 2.52 .69 3.40 .78 −1.11 Erudition (.75) 2.84 .68 2.91 .68 2.69 .66 .29 Note. Community N = 379 women, 286 men; college N = 245 women, 101 men. Internal-consistency reliabilities are in parentheses. ∗ ns; for all other d values, p
Page 1: Journal of Personality Assessment,
Page 5 and 6: 172 POZZEBON ET AL. TABLE 4.—Corr
Page 7 and 8: 174 POZZEBON ET AL. Adventure (CISS

Pozzebon et al.,2010

Create successful ePaper yourself

Delete template?

Save as template?